Robustness checks: gender quotas and glass cliff: fixed effects panel regression models (2SLS)
| Model 1 Roe as endogenous variable | Model 2 Roa as endogenous variable | Model 3 Bad performance as endogenous variable | ||||
|---|---|---|---|---|---|---|
| Variables | First stage Y = Roe | Second stage Y = WoBs | First stage Y = Roa | Second stage Y = WoBs | First stage Y = Bad performance | Second stage Y = WoBs |
| Board size | −0.01 (0.021) | −0.01 (0.008) | −0.001 (0.001) | −0.02 (0.015) | 0.01 (0.016) | −0.01 (0.008) |
| Firm size | 0.03 (0.046) | 0.10 (0.071) | −0.003 (0.008) | −0.003 (0.113) | −0.132 (0.100) | 0.13** (0.062) |
| Law | −0.01 (0.021) | 0.13*** (0.036) | −0.00 (0.004) | 0.14** (0.073) | 0.02 (0.051) | 0.13*** (0.032) |
| Women executives | 0.08* (0.048) | 0.26*** (0.082) | 0.01 (0.008) | 0.26** (0.142) | −0.06 (0.113) | 0.18*** (0.066) |
| Roe | −1.48*** (0.333) | |||||
| Bad performance | 0.55*** (0.100) | |||||
| Roa | −15.99** (6.862) | |||||
| Asset-to-equity ratio | 0.00*** (0.000) | 0.00*** (0.000) | −0.00*** (0.000) | |||
| Sanderson-–Windmeijer F test of excluded instruments | F (1,111) = 16.42; Prob > F = 0.0001 | F (1,111) = 4.92; Prob > F = 0.028 | F (1,111) = 30.58; Prob > F = 0.000 | |||
| Observations | 232 | 232 | 232 | 232 | 232 | 232 |
| F – test; Prob > F | F (5, 111) = 17.18; p = 0.000 | F (5, 111) = 26.12; p = 0.000 | F (5, 111) = 55.33, p = 0.000 | |||
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Variables | ||||||
| −0.01 (0.021) | −0.01 (0.008) | −0.001 (0.001) | −0.02 (0.015) | 0.01 (0.016) | −0.01 (0.008) | |
| 0.03 (0.046) | 0.10 (0.071) | −0.003 (0.008) | −0.003 (0.113) | −0.132 (0.100) | 0.13 | |
| −0.01 (0.021) | 0.13 | −0.00 (0.004) | 0.14 | 0.02 (0.051) | 0.13 | |
| 0.08 | 0.26 | 0.01 (0.008) | 0.26 | −0.06 (0.113) | 0.18 | |
| −1.48 | ||||||
| 0.55 | ||||||
| −15.99 | ||||||
| 0.00 | 0.00 | −0.00 | ||||
| F (1,111) = 4.92; | F (1,111) = 30.58; | |||||
| 232 | 232 | 232 | 232 | 232 | 232 | |
| F (5, 111) = 55.33, | ||||||
Notes:
Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1